Data Package Metadata   View Summary

Data Source & Collection

We conducted an analysis using macroinvertebrate taxon abundance data from 1060 stream reaches sampled by the Maryland Biological Stream Survey (MBSS; https://dnr.maryland.gov/streams/pages/mbss.aspx). The MBSS is a stream-monitoring program based on probabilistic sampling stratified by major basins and stream order (first to third order on a 1:250 000 stream map during 1995–1997 and first to fourth order on a 1:100 000 stream map during 2000–2003). Sampling locations were randomly assigned to non-overlapping, 75-m stream segments across the state of Maryland. Within each randomly selected segment, the survey included rapid- assessment metrics of stream physical characteristics, conventional water chemistry, in-stream habitat, discharge, and macroinvertebrate assemblage composition (2-m2 sample collected with a 500-um D-framed dip net from favorable habitat, genus-level identification, 100- organism, fixed-count method. All of the selected sites were sampled once during 1995–2006, a period that overlapped with the 1985–2010 satellite imagery used to quantify impervious cover (see below).

Macroinvertebrate data were screened for consistency prior to analysis and ambiguous taxa were resolved following the operational- taxonomic-unit-approach. If more than 100 individuals were enumerated for a stream reach, we used a computer- generated random sampling procedure to select of subsample of 100 individuals to ensure that individual taxa frequencies and abundances were expressed consistently for all sample units because both depend on the number of individuals in the subsample. Samples with fewer than 100 individuals were not altered because they represent stream reaches with low densities of individuals possibly due to anthropogenic stressors, including catchment imperviousness.

Phylogenetic Tree Construction

A list of target taxa was created by selecting those from the master data set designated as shredders by the MBSS. Only insect taxa were included in the analysis. We used the command-line program phyloGenerator to create a phylogenetic tree. The method for the creation of the this tree is based on the BioPython framework and automates the retrieval of sequence information from Genbank, in addition to automating alignment and phylogenetic tree estimation. We constructed our tree based on five commonly sequenced genes, COI, 16S, 18S, 28S, and H3. The trimmed sequences were aligned using MAFFT after being visually examined for misalignment using ClustalX. We then generated a maximum likelihood tree with RaxML, using 1000 accelerated bootstrap replicates and constrained the final tree based on a taxonomic tree. Finally, we rate-smoothed the resulting tree to be ultrametric using the PATHd8 algorithm, with the Apis genus set as an outgroup.

Phylogenetic Diversity

This was calculated as the total branch length of all species observed for each assemblage. We calculated the abundance-weighted mean nearest taxon distance (MNTD). MNTD measures the mean distance between each species’ closest relative in a community, and thus focuses on similarity between co-occurring species. The number of species in a community and the structure of the phylogeny of the regional species pool can influence both metrics. To remove effects of species richness and regional phylogenetic structure, we compared observed MNTD against metrics calculated from null communities. We generated null communities by randomly drawing the same number of species as observed at each site from the regional species pool. Null communities were assembled by randomly drawing N species from the pool, where N was the number of species observed at a given site. The probability of drawing a species was proportional to the frequency that the species occurred within transects that defined the species pool. Next, we calculated MNTD for all 1000 null community sites, and expressed our results as the standardized effect site (z).

Taxonomic Diversity

Taxonomic diversity here was taken simply as taxa richness, or the number of taxa at each site.

Mean Phylogenetic Distinctiveness

We quantified the evolutionary distinctness of each species using the fair proportion metric, then followed with applying the value for each species in an assemblage. The average for the assemblage was taken and then related to impervious cover.

Analysis of Impervious Cover

Impervious cover was estimated from a time series of Landsat satellite imagery using the Dynamic Impervious Surface Cover Observation and Retrieval System (DISCORS). DISCORS uses high-resolution, municipal planimetric data collected over time to train a nonparametric regression tree, which is then applied to both wintertime reflectance and changes in summertime greenness to estimate impervious cover at annual, 30-m resolution across the study area from 1985 to 2010. Only one year in the series, 2009, was interpolated from 2008 and 2010 due to insufficient cloud-free satellite measurements. Pixels in the raster are zero (no impervious cover) or have an integer value of percent imperviousness ranging from 10% to 100%. DISCORS produced a 25-yr time series of annual impervious cover for the Baltimore-Washington, DC metropolitan area with misclassification error (i.e., ~6% per pixel RMSE) comparable to the impervious cover layer of the National Land Cover Database (NLCD). We distinguished real interannual change from noise by ignoring annual per-pixel differences <12% (i.e., RMSE*2). Aggregation of pixels to the scale of catchments (typically thousands of pixels) neutralizes omissions and commissions due to randomly distributed errors, so catchment-scale impervious estimates became increasingly precise as a function of increasing spatial scale.

Contributing areas for each stream sampling site were delineated using the 1/3 arc-second DEM resolution, which is approximately the 10m resolution obtained from the National 3D Elevation Program (3DEP) https://www.sciencebase.gov/catalog/item/4f70aa9fe4b058caae3f8de5. DEMs were hydrologically corrected by filling internal drainages, deriving a flow direction field, and accumulating drainage. The contributing area upstream from each sample point was delineated following its manual alignment with the closest flow line and was used to summarize the proportion of impervious cover for each year in the impervious time series.